Defining the Speaking Threshold Kernel (Learning Techniques, Methods & Strategies) Language Learning Forum

Defining the Speaking Threshold Kernel
Tags: Number of words \| Speaking
Share with: Delicious Digg reddit Facebook StumbleUpon
Language Learning Forum : Learning Techniques, Methods & Strategies

28 messages over 4 pages: 1 2 3 4 Next >>

robarb
Nonaglot
Senior Member
United States
languagenpluson
Joined 4855 days ago
361 posts - 921 votes

Speaks: Portuguese, English*, German, Italian, Spanish, Dutch, Swedish, Esperanto, French
Studies: Mandarin, Danish, Russian, Norwegian, Cantonese, Japanese, Korean, Polish, Greek, Latin, Nepali, Modern Hebrew

Message 17 of 28

14 October 2014 at 5:35am | IP Logged

s_allard wrote:

OK, maybe the word "dodgy" was a bit too strong for what I perceived as a contrived conversation that seemed
unnatural to me. I take back my claim that this was done for purposes of pointing out missing items. That said, I
really don't see the point of making up examples where there is so much real stuff.

That's exactly why I like to work with transcriptions of real conversations. Rather than trying to invent things, I
like to work with transcriptions of natural French conversations for two reasons. First, they illustrate how the
language is really used. Everything is there, including, as I pointed out, all the features that are typical of spoken
language plus the recordings with the phonetic component.

Second, the transcripts are very useful for purposes of validation. I map the kernel list with its underlying
linguistic knowledge against the transcripts and I see a good fit. I see lots of holes for content nouns, of course,
but that's not a big problem. For example, in the following sequence,

A : Alors Philippe, vous vous êtes levé à quelle heure ce matin ?
P : Donc ce matin, je me suis levé à six heures et quart, pour commencer à huit heures.
A : Oui. Donc ça veut dire quoi, là ? Vous partez à quelle heure de chez vous ?
P : Je pars vers 6h55 – 7h de chez moi.

Not only is nearly every single word from the list, but I also see the underlying grammar and particularly natural
phrasing. A perfect combination of words and linguistic knowledge.

From a learner's perspective, the question is: What does it take to be able to talk like this?

Fair enough. If you're committed to using natural conversation and not artificial constructed ones, let me suggest
a way to make a much clearer demonstration than simply posting snippets of natural conversation:

1. Take a relatively large sample of a simple, natural conversation. It should be at least 20 turns.
2. Identify the words in the dialogue that aren't in the kernel.
3. Rewrite the conversation, replacing those words with alternatives contained within the kernel.
4. If at any point step (3) runs into difficulties, describe the repair strategies (gesture, saying something simpler)
that would be appropriate for a learner in this situation.

If you rarely have to use step (4), then you've shown that the kernel is enough to handle the situation, and a
learner with just this vocabulary can function autonomously. This seems likely to be the case for conversations
similar to what you've posted above. Perhaps for some more difficult conversations (maybe something abstract,
technical, political, or involving several specific uncommon items), step (4) would frequently be needed. In that
case, a learner with this level of vocabulary would not be able to function as a fully autonomous speaker in that
situation. They can by all means still try to speak in that situation, assuming a repair channel is available (e.g. the
other speaker knows the learner's L1) and efficient communication isn't critical.

s_allard wrote:

One could also discard the whole idea of a threshold and say that there is simply a range from 48 (four dozen) to
1000 for the learner to start speaking.

I think this is correct. The number of words is always going to depend on the range of situations you want to be
able to function in as an autonomous speaker. If that's exclusively greetings, directions, and buying, maybe it's
48. If it also includes small talk, scheduling, personal needs, politeness, self-introduction, maybe it's 300. If you
also include ethics, language learning methods, biology, music, political debate, telling someone what groceries
to buy, job interviews, and telling jokes, and quarreling, maybe it's 1000. For any range of situations you want to
be able to handle, there's an approximate threshold of core vocabulary. I believe it's 1000 or less for the set of
topics you'd expect a person to speak about without them being an expert. The level of talkativeness you want
and the frequency of failures that's acceptable make the actual numbers subjective. Native speakers use more
than 1000, but after 1000 the additional words are almost never required to get the basic idea across.

1 person has voted this message useful

s_allard
Triglot
Senior Member
Canada
Joined 5226 days ago
2704 posts - 5425 votes

Speaks: French*, English, Spanish
Studies: Polish

Message 18 of 28

14 October 2014 at 7:28am | IP Logged

I take robarb's suggestion under advisement. I really don't have the time to do such an experiment, and I'm not
sure how relevant it is. Plus I'm much more interested in Spanish right now than in French.

But, more importantly, I think that for what it purports to do, the suggested 300-word kernel seems fine to me.
Let me remind everyone that the idea is not to suggest that one can debate a wide range of subjects in depth
with this kernel. I won't quote myself but, as outlined in the OP, the fundamental idea is that, given a certain
sum of linguistic knowledge in the target language, the learner can start having relatively simple interactions on a
simple range of topics with native speakers.

Why 300 units and not 48 or 1000? There are two different issues here. First of all, is the idea of a threshold
useful at all? Is there a kind of tipping point where the learner can start to feel a level of confidence using the
language for real? I believe there is, and I think it corresponds to something that most people experience, not
necessarily as a sharp demarcation but certainly as a feeling that they are beginning to communicate in the
target language.

Secondly, if one believes in a threshold, can one put a word-count figure on this? Again, as I pointed out in the
OP, we feel that this figure is a very imperfect metric for what is really linguistic knowledge and much more than
words. For example, the depth of knowledge or mastery of the kernel can vary considerably and this will
determine the actual speaking proficiency.

For example, I see absolutely no problem for someone to go into a supermarket, ask for the location of a
product, find it and pay at the checkout counter, There are all sorts of potential complications, as is often
pointed out. Where I live, customers will be asked if they have a fidelity card such a Air Miles, Aeroplan,
Optimum, etc. or if they want a bag. One can pay by credit card, bank card or cash. Some supermarkets even
have self-serve checkouts where the interaction is with a machine. The 301-word list I gave does not have the
words for the fidelity cards, but I don't see why this should be a problem if one is observant while waiting in line.

With this list, one should be able to buy some tickets to a concert and pay with a credit card. Now, if the credit
card is declined, there could be a messy situation where your limited linguistic repertoire or your lack of
linguistic mastery may fail you, but barring various rare complications, this sort of linguistic transaction should
not be a problem.

The problem most people see with this idea of a kernel is lack of words. I would like to add that the depth of
linguistic knowledge is just as important even though it's not as easy to measure. The complexities of
preposition or pronoun usage in French are many. The interference of the first language is an important issue.
Good phonology makes a huge difference in intelligibility.

That is all the difference between a native speaker with these 300 words in a real-life situation and our learner
with a very imperfect mastery of the language and the interfering presence of a first language.

There are many things one cannot do with a 300-word repertoire, even with maximum linguistic mastery. That's
why you need more words, especially nouns. We all agree on that. I still like the figure 300 for the kernel
because I think it covers the essentials, but if someone believes that 1000 is a better, I really don't have a
problem with that figure. 48 is a bit of a stretch, however.

Edited by s_allard on 14 October 2014 at 7:33am
1 person has voted this message useful

Jeffers
Senior Member
United Kingdom
Joined 4705 days ago
2151 posts - 3960 votes

Speaks: English*
Studies: Hindi, Ancient Greek, French, Sanskrit, German

Message 19 of 28

14 October 2014 at 1:29pm | IP Logged

s_allard wrote:

Why the idea of threshold if people can start speaking from day 1. Isn't Bonjour, ça va? and Merci beaucoup speaking? Couldn't one start speaking French with "a couple dozen words"? I'm glad that I'm not the one saying that you can speak French with a few dozen words.

The idea of a threshold is that around a certain number of words, e.g. 300, the learner is somewhat autonomous and can cover a wide range of situations beyond ordering a coffee in a café. 250 could also be a threshold. Maybe 400 is better. 300 is a figure that one finds in the literature, but it's arbitrary.

One could also discard the whole idea of a threshold and say that there is simply a range from 48 (four dozen) to 1000 for the learner to start speaking.

I quite like the way you've described it here. The threshold is a point a learner can work towards from which they can start becoming "autonomous", in other words a bit more independent. But as you say, it's more of a spectrum than a threshold. My worry is that the idea of a threshold would be taken by a lot of people as an end in itself, and they would be disappointed that nothing magically changed on the day they mastered word 300. (I know this is nothing you have ever claimed, although "tipping point" does come close. Anyway, a lot of learners are looking for a silver bullet and believe there is some sort of magic or voodoo in a particular method).

Edited by Jeffers on 14 October 2014 at 1:32pm
1 person has voted this message useful

s_allard
Triglot
Senior Member
Canada
Joined 5226 days ago
2704 posts - 5425 votes

Speaks: French*, English, Spanish
Studies: Polish

Message 20 of 28

14 October 2014 at 2:26pm | IP Logged

At this very moment, there are two interesting language-learning experiments going on that can illustrate what
the speaking threshold idea is about.

A fellow HTLALer and very experienced language learner is attempting to passively learn Mandarin by watching
videos 30 minutes a day.
Learn exclusively with authentic
videos

I mention this experiment because it is the polar opposite of every thing that I have been discussing in this idea
of a speaking threshold. There is no interest in speaking - for the time being - and absolutely no sense of
direction.

The second experiment, as mentioned by HTLALer Glarus Girl, is an attempt by a well-known language teacher,
Michael Schmitz to learn modern Greek to B1 level in 30 days using his own teaching/learning techniques. This
experiment will end around October 21.

Greek in 30
days

I haven't been following the daily video blog of Schmitz's progress but from what I see, he is going about it the
right way, using Assimil and daily sessions with a private tutor. It's great to see a professional language learner
at work.

I believe that Schmitz is implicitly using the threshold idea - as I've said, it's nothing really that new. Suppose
that Schmitz started out, under the guidance of his tutor, with a formal threshold list like the one we have here
and mapped it to Assimil and the private sessions. I think the results would be even more spectacular. Not that
Schmitz needs any advice from me.

All these very intensive language learning courses work pretty much the same way. They saturate the learner
linguistically and try to introduce enough fun and entertainment so that the student is not totally overwhelmed.

Here are the end results for a student of French after seven weeks and $10,000 at Middlebury College in the US:

After seven
weeks of French

I should point out that the speaker makes a number of speaking mistakes i- what I would call lacking in
linguistic knowledge - but it does show what can be done in seven weeks.

Edited by s_allard on 14 October 2014 at 2:28pm
1 person has voted this message useful

Serpent
Octoglot
Senior Member
Russian Federation
serpent-849.livejour
Joined 6393 days ago
9753 posts - 15779 votes

4 sounds
Speaks: Russian*, English, Finnish^C1, Latin, German, Italian, Spanish, Portuguese
Studies: Danish, Romanian, Polish, Belarusian, Ukrainian, Croatian, Slovenian, Catalan, Czech, Galician, Dutch, Swedish

Message 21 of 28

14 October 2014 at 2:27pm | IP Logged

s_allard wrote:

Now, if the credit card is declined, there could be a messy situation where your limited linguistic repertoire or your lack of linguistic mastery may fail you, but barring various rare complications, this sort of linguistic transaction should not be a problem.

B1 does assume being able to cope with such problems though.
1 person has voted this message useful

jeff_lindqvist
Diglot
Moderator
SwedenRegistered users can see my Skype Name
Joined 6705 days ago
4250 posts - 5710 votes

Speaks: Swedish*, English
Studies: German, Spanish, Russian, Dutch, Mandarin, Esperanto, Irish, French
Personal Language Map

Message 22 of 28

14 October 2014 at 7:10pm | IP Logged

Tipping point, threshold, kernel, spectrum - some learners believe in exact numbers. Many of us have heard about the 10 000 hours, and there have been recent threads about the TV method (300 hours? 1000? 2000?).

Personally, I'm convinced that I won't realize when I've passed the imagined tipping point until long after I really passed it. If it's ~300 words, I won't notice that until my vocabulary is in the thousands.
4 persons have voted this message useful

s_allard
Triglot
Senior Member
Canada
Joined 5226 days ago
2704 posts - 5425 votes

Speaks: French*, English, Spanish
Studies: Polish

Message 23 of 28

21 October 2014 at 11:31pm | IP Logged

For those readers who may have an interesting in speaking French, and for some following posts, here is a
revised version of the Speaking French Threshold Kernel. There's been a fair amount of tweaking with the various
categories.

Version 1.1 of a French 301-word Speaking Threshold Core

Breakdown by functional category:
Verbs     59
Nouns     118
Adverbs     14
Adjectives 46
Pronouns 24
Connecting words 40

Total     301

Verbs wrote:

acheter, aider, aimer, aller, arriver, atteindre, avoir, boire,     changer, chercher,
commander,commencer,comprendre,connaître,continuer, coucher,coûter,
démontrer,dépendre,devenir,devoir, dire,donner,dormir,durer,
éclater,écouter,entendre,entraîner,essayer,être,explique r, faire,falloir,
finir,habiter,imaginer,intéresser,laisser,laver,manger, naître,
passer,peser,porter,préférer,prendre,préparer,prier,profi ter,
raconter,regarder,rester,savoir,tenter,trouver,varier,vivre, voir,vouloir

Nouns wrote:

la baguette, la campagne,la carte,la chambre,la chance,la chanson, la chaussure,la chemise,la cuisine,la
discipline,la face,la famille,la fille, la fois,la forme,la gare,la maison,la mère,la nouvelle,la passion,la peau,la
personne, la qualité,la raison,la réservation,la robe,la salle,la soeur,la télévision,la tête,la vacance,la ville,la voie,
la voiture,l'accident,l'an, l'année, l'argent,l'autobus,l'aventure,le bain,le besoin,le billet,le bien,le café,le cas,le
centre,le chapeau,le charme,le cinéma,le cours,le crédit,le début,le déjeuner,le dimanche,le dîner,le fils,le frère,le
fruit,le gramme,le haut,le jeudi, le jour,le jus,le kilo,le lit,le livre,le lundi,le magasin,le mal,le manteau,le
mardi,le matin,le menu, le mercredi,le métier,le métro,le midi,le nord,l'examen, le pantalon,le parent, le père,le
peu,le problème,le professeur,le progrès,le quartier,le rapport,le rendez-vous,le repas,le restaurant,le sac,le
salon,le samedi,le sport,le sud,le tarif,le tort,le tout,le train,le travail,le type,le midi,le voyage,le vélo,le visage,le
soir,le vendredi,le vêtement,l'eau,l'enfant,l'est, l'étude,l'exemple, l'expérience,
l'extérieur,l'intérieur,l'ouest

Adjectives wrote:

absolu,accessible,autre,cent,chaud,cher,cinq,cinquante,clair ,dernier,deux,deuxième,dix,douze,efficace,faux,fiabl
e,froid,grand, huit,neuf,onze,
particulier,petit,premier,quarante,quatorze,quatre,quatrièm e,quelque,quinze,récent,seize,sept,six,   &n bsp;
soixante,ton/ta/tes,treize,trente,trois,troisième,un,vieux, vingt,vrai

Adverbs wrote:

assez,
beaucoup, bien, finalement,ici,jamais,là,la-bas,la-haut,longtemps,mal,notam ment,toujours, très, vite,vraiment

Pronouns wrote:

auquel, ça,ce,ceci,celui,elle, en,eux,il, je,la,le,lequel,lui, moi,nous,on,quelqu'un,soi,te,toi,tu,vous,y

Connecting words wrote:

à, à côté, alors, après, au revoir, avant,bonjour,chez,     combien,com me,comment,d'accord,dans,de,     de puis,
donc,enfin,ensuite,et,hein,la,      là,le,mais,même,ne pas,non,ou,oui,plus,puis,quand,qu'est-ce
que,quoi,si,vers,voici,voilà

1 person has voted this message useful

s_allard
Triglot
Senior Member
Canada
Joined 5226 days ago
2704 posts - 5425 votes

Speaks: French*, English, Spanish
Studies: Polish

Message 24 of 28

22 October 2014 at 12:33am | IP Logged

robarb wrote:

...

Fair enough. If you're committed to using natural conversation and not artificial constructed ones, let me suggest
a way to make a much clearer demonstration than simply posting snippets of natural conversation:

1. Take a relatively large sample of a simple, natural conversation. It should be at least 20 turns.
2. Identify the words in the dialogue that aren't in the kernel.
3. Rewrite the conversation, replacing those words with alternatives contained within the kernel.
4. If at any point step (3) runs into difficulties, describe the repair strategies (gesture, saying something simpler)
that would be appropriate for a learner in this situation.

If you rarely have to use step (4), then you've shown that the kernel is enough to handle the situation, and a
learner with just this vocabulary can function autonomously. This seems likely to be the case for conversations
similar to what you've posted above. Perhaps for some more difficult conversations (maybe something abstract,
technical, political, or involving several specific uncommon items), step (4) would frequently be needed. In that
case, a learner with this level of vocabulary would not be able to function as a fully autonomous speaker in that
situation. They can by all means still try to speak in that situation, assuming a repair channel is available (e.g. the
other speaker knows the learner's L1) and efficient communication isn't critical.

...

One of the issues discussed in our working group is how to verify the usefulness of this idea of the speaking
threshold core. The most obvious solution would be to see how the core matches up in terms of coverage of a
range of real conversations, something along the lines of what robarb has suggested. I objected that since most
of the sample conversations have at least two speakers, if we are looking at the ability to participate in a
conversation, we have to look at the speech of only one given speaker.

This speaker must be able to deal with the vocabulary of the other speakers, of course, but that is a separate -
but important - problem of receptive vocabulary. I would argue that in terms of actual speaking performance,
this is a more realistic model. For example, looking from the perspective of the person doing the interviewing,
the following dialogue from France Bienvenue
La patience de
Mélissa
becomes:

A : Mais vous l’aviez eu comment, ce travail ?
 …
 A : Oui ?
 …
 A : Oui.
 …
 A : Vous avez cherché…
 …
 A : Oui.
 …
 A : Oui. Bah j’imagine que… ils doivent être contents de vous avoir parce que…
 …
 A : … si vous travaillez aussi bien là-bas qu’ici…
 …
 A : Vous êtes la perle !
 …
 A : Oui, oui. Vous êtes fiable.
 …
 A : Oui, j’imagine.
 …
 A : Oui, c’est pas le cas de tout le monde, hein !
 …
 A : Et les clients ?
 …
 A : C’est vrai ?
 …
 A : Il y a de tout ?
 …
 A : Patients.
 …
 A : Oui ?
 …
 A : C’est vrai ?
 …
 A : Et alors ? Il faut rester calme ?
 …
 A : Eh oui.
 …
 A : C'est vrai ?
 …
 A : A ce point ?
 …
 A : Ouh là, là !
 …
 A : Parce qu’ils sont pressés, ça va pas assez vite.
 …
 A : Ah oui !
 …
 A : Et ils s’en prennent à vous.
 …
 A : D’accord.
 …
 A : C’est un peu dur, ça, quand même, non ?
 …
 A : Vous arrivez à faire un peu abstraction, après, ou ça perturbe ?
 …
 A : Bah oui. Très pacifique.
 …
 A : Très polie, très courtoise.
 …
 A : Oui, oui. J’imagine.
 …
 A : C’est sûr.
 …
 A : Oui ?
 …
 A : Bah c’est sûr, hein.
 …
 A : Oui, on reste enfant quand on reste…
 …
 A : Enfin, on reste un peu enfant quand on est…
 …
 A : Oui, oui. C’est bien !
 …
 A : Très bien.
 …
 A: Bah merci beaucoup, Mélissa !
 ...

A number of words in this speaker's output are not in the 301-word core: Nouns: les clients, l'abstraction
Adjectives:patient,calme;
pressé, pacifique, polie, courtoise; Verb: perturber.

Other than those words, everything else is in the core. The fit is, in my opinion, very good.

Let's look at another conversation using this same idea of a unique speaker's perspective. Here is the original
conversation:
La préférée du roi

Here is output of the speaker doing the interviewing.

A : Donc bonjour Shainesse .
 …
 A : ça va bien?
 …
 A : La semaine démarre bien?
 …
 A : Oui ? beaucoup de boulot ? Ça y est ?
 …
 A : D’accord. Donc en fait, bon, je vous connais puisque qu’on était ensemble en première année, mais je vous
connais sans vous connaître.
Alors je voulais savoir un petit peu… Bon, vous vous appelez Shainesse, et c’est… je me dis, bon, quelle est votre
origine exactement?
 …
 A : C’est vrai ?
 …
 A : Et pourquoi elle avait choisi alors ce prénom-là, perse ?
 … A : Ah d’accord!
 …
 A : D’accord, parce que votre pays, c’est quoi? C’est pas la France?
 …
 A : Alors les Comores, mais ça veut dire que vous y avez… vous y êtes née, vous y avez grandi ?
 …
 A : D’accord. Mais avec la famille ou toute seule?
 …
 A : Et votre maman, elle est aux Comores?
 …
 A : D’accord. Donc quand vous êtes partie à 14 ans, ça a fait quand même… Enfin, ça devait être une grosse,
enfin un gros changement?
 …
 A : … qui étaient déjà bien intégrés, bien habitués. Mais c’était à Marseille? Ils étaient installés à Marseille ?
 …
 A : D’accord. Et comment vous l’avez vécu ça, alors quand même? Parce que vous dites que ça a un peu amorti le
choc et tout. C’était dur
quand même?
 …
 A : Le climat, tout ça ?
 …
 A : Il faisait jour.
 …
 A : D’accord, donc des petites choses comme ça qui surprennent vraiment.
 …
 A : Oui. Vous êtes arrivée en quelle classe alors?
 …
 A : D’accord. Et c’était facile de s’intégrer, enfin au niveau du… par exemple des programmes, tout ça ? C’était…
Vous vous êtes pas sentie à la
traîne ou…?
 …
 A : Non pas du tout ?
 … A : Et il y a des liens importants entre la France et les Comores.
 …
 A : Oui, pas dépaysant.
 …
 A : Et en fait, vous avez toujours parlé français ?
 …
 A : Ou on parle autre chose aux Comores?
 …
 A : Vous parlez les deux?
 …
 A : Bon bah c’est intéressant tout ça, c’est une belle expérience.

In this conversation, the fit of the core words is not as good as in the first conversation. Here are the words
not in the 301-word core: Nouns: la chose, le boulot, l'année, le français, le choc, la classe, le lien, le
programme, la maman, la traîne, l'origine, le prénom; Verbs: s'intégrer, surprendre, grandir, amortir ;
Adjectives:installé, dépaysant, perse Connecting Words: puisque. One could also add proper nouns
such as: la France, les Comores, Marseille.

These results are pretty much what I expected. The core words and grammatical components cover the vast
majority of the speakers' output. Just in terms of grammar there are no surprises. The missing words are to be
found in the noun category, as to be expected.

I'm not disregarding the issue of receptive vocabulary here, but just in terms of actually speaking we see that the
300-word threshold is a pretty good starting point for speaking French. If we bump that number up to let's say
400 by including a few more verbs and a significant number of nouns, then speaking on a wider range of topics
will become much easier.

Edited by s_allard on 22 October 2014 at 1:49pm

1 person has voted this message useful

This discussion contains 28 messages over 4 pages: << Prev 1 2 3 4 Next >>

Printable version

You cannot post new topics in this forum - You cannot reply to topics in this forum - You cannot delete your posts in this forum
You cannot edit your posts in this forum - You cannot create polls in this forum - You cannot vote in polls in this forum

This page was generated in 0.3926 seconds.